Editor’s Note: Retraction of article containing fabricated quotations

We are reinforcing our editorial standards following this incident.

See full article...

 

[Sarty]

This is one of those perfect demonstrations of how AI does not save time. In this case:
1. It did not save time summarizing the article because the first tool failed to work at all.
2. It did not save time because the second tool generated a summary filled with quotes that never existed.
3. It did not save time because the person the article was about had to come and fact check the article himself in the comments.
4. It did not save time because multiple members of the Ars staff had to look into the situation on a weekend.
5. It did not save time because Kyle had to post about his innocence.
6. It did not save time because Benj had to post his explanation and apology.

Many excellent comments and viewpoints about this Charlie Foxtrot, but it took until page 9 to get the best one.

 

[Sarty]

It's very sad to see Mr. Edwards seemingly throw away his career like this, but that would be the only reasonable and acceptable outcome. "Sorry I was driving tired when I drove my bus over a crosswalk full of schoolchildren". Well... yes, but passersby were amazed by the unusually large amounts of blood. Passersby were amazed by the unusually large amounts of blood. Passersby were amazed by the unusually large amounts of blood.

Sometimes you walk old battlefields or ancient European cities and see intentionally-unrepaired bullet holes in columns and monuments. "We left it like this so we wouldn't forget". I hope that's the long-term outcome for Ars Technica--made stronger by a moment of weakness.

 

[Sarty]

You want what, a public execution?

Contingent on employment law and contractual details, it would be absolutely appropriate for Ars Technica to announce "We have fired our senior AI reporter for an egregious violation of policy and professional ethics in misusing an AI tool to generate false reporting".

Frankly, any lesser response will be very worrisome.

 

[Sarty]

No they don't. But at the same time, this was a systemic failure. Benj Edwards very clearly violated existing policy, something that a senior writer (covering the AI beat no less) should never, ever do.

But he's not alone, apparently whatever review articles are put through before they're published is woefully insufficient.

One aspect of sympathy I'll extend to Ars is that you really do need to be able to expect that your employees are, on the whole, acting in generally good faith. You should have measures in place to catch somebody stealing from the corporate coffers, stuff like that, but it's hard to imagine a system operating with any kind of efficiency when you have to assume everyone is acting in bad faith, all the time. This is even a who-watches-the-watchers situation--Mr. Edwards was not junior staff. He ostensibly runs (hopefully ran) this beat.

AI slopboxes can't act in good faith. They can't act in any faith at all. In the journalism space, I think the only reasonable policy is zero tolerance for usage, and demonstrated harsh outcomes for those who violate that policy. If you want to vibecode flappy bird 2.0, maybe that's not my problem.

It keeps me up at night. I have enormous professional respect for, and personal trust in, my employees, but they are relentlessly bombarded by stories about AI magic this and that--from mass media, from popular culture, even from my own bosses.

 

[Sarty]

if it is verified as an isolated occurrence about which everyone is honest within a short, but non-zero, amount of time

a.k.a., "I only stole from the company once"

Once is fucking enough!

 

[Sarty]

While I get the point you're trying to make, comparing this to vehicular manslaughter (of children no less) is way off base.

A journalist fabricating quotes from a real person (or fabricating evidence more broadly) is nearly the worst non-malicious professional act I can imagine within their field, just as accidentally running over a bunch of children is nearly the worst non-malicious professional act a bus driver could perform. The implied equivalence was completely intentional and I stand by it confidently and resolutely.

 

[Sarty]

But as a human being, I also hate the idea that a screw-up while sick, which was likely a quick "this will only take a minute and nobody will know," has thoroughly fucked Edwards's job and reputation. He should have known better - he's clearly spent a lot of time around these abominations, as his job, and should be well aware of their pitfalls. But I feel sorry for him, and yeah, I hope he has a chance at professional redemption, either here or elsewhere.

It would be much sadder if 20 (?) Ars Technica journalists lost their jobs because nobody had a whisker of faith in the journalistic standards of the organization and the whole thing went under. Or the other writers at any other organization that foolishly hired him into a like role after this.

It's very regrettable, but only one actor here made his bed.

 

[Sarty]

5. Shit hit the fan, etc. As I understand it, there was an active comments section following Shambaugh's engagement.

If it's interesting context or if anybody cares regarding "active comments section", that comment section ran for about two pages (definitely sure) in maybe ninety minutes (less sure).

 

[Sarty]

I don't think 5a happened before 6. Listing it first implies Ars tried to shut down the conversation while leaving the article up, and I don't know that it actually happened that way. Of course once the original article was pulled, it makes sense that the comments would also be hidden, that has happened before.

The comments were locked somewhere around 3:30 Central (quoting from my own memory here), and the article remained up for, eh, a few tens of minutes thereafter. But the duration was short enough that it could well have been the time it takes between clicking one button and clicking another.

 

[Sarty]

I'm on the fence about zero tolerance for using AI in this line of work. First, it's completely unenforceable for remote work. Second, we don't know what the future holds for this tech and its perception. Third, I think the best policy is to have those that submit work be accountable for what they submit, regardless of how it is produced.

1) I think enforceability is a bit of a canard. We have all kinds of policies that would be difficult or impossible to enforce in a rigorous manner. Hell, the fast food joint down the street does not have a camera in the bathroom ensuring that employees wash their hands after relieving themselves. Nevertheless, if we have a policy that says "You shall not do ABC and this is the reason why", a lot of conscientious employees will follow that policy regardless of whether they think they'll get caught. Which matters because

2) If this sad episode has made anything clear, it is that no matter how plugged into the AI landscape you are, no matter how aware you are of its pitfalls, your squishy meat brain is desperate to trust the magic slop box and treat it as though it were a careful fellow human. It is irresistible. It is vitally important to deter this bad behavior. A no-use policy is designed to save you from yourself. It could have, should have, saved Mr. Edwards' career in journalism.

 

[Sarty]

This is part of why I flatly refuse to use LLM slopcoding machines in my work.

I will fail to check something well enough at some point. I don't want to deal with the consequences of a slopcoded test I failed to properly vet letting a bug through and potentially fucking up processing a card transaction or thousand.

It's a drug, and the first one's free.

I absolutely don't think I'm resistant to falling for the LLM siren, not any more than the next dumb guy. That's precisely why I do not use them and am not interested in starting to use them.

You might as well fry your brain-eggs on Facebook or TikTok.

 

[Sarty]

Unemployment benefits generally are not available for an employee resigning voluntarily.

I think Jim's point is that Mr. Edwards does not face the question of "resign or be downsized", where there is an open question of what post-employment resources may be available. He perhaps faces the question of "resign or be fired for cause", where there is no question anyway--zero versus zero.

 

[Sarty]

Where the gamesmanship comes in is trying to figure out whether you'll get let go or fired, because if it's the former, you don't want to resign and lose your benefits--but if it's the latter, you REALLY want to resign first, so you don't have to put a termination for cause on your resume (or lie about it).

Of course, that presumes you're in a semi-anonymous position where hiding is even remotely possible.

It feels a little crass to talk like this while someone still has what I assume is a walking-dead position in the organization, but if I were a friend, I would strongly advise the corresponding author to submit his resignation effective immediately. Secondarily and much less strongly, offer to talk with another Ars journalist and try to put together a very frank "this is how this can all go wrong" article under the latter's byline. That is--as a source, not a fellow author. That plus whatever handoff/cleanup work to be billed at a reasonable hourly rate, as seen fit by management.

 

[Sarty]

Did you read the article you linked, cause I've read the first dozen or so paragraphs and it doesn't support the point you're making.

No, but he had a LLM pluck a couple of select quotes* out of it.

 

[Sarty]

That's why Ars buried it as hard as they could, then when they lost containment they recreated the article (rather than un-unpublishing it) or deleted all comments on it, don't state who did the thing, what the thing they did was, and otherwise assign no actual accountability.
...
Ars got caught aiming that firehose at their audience, lost containment of the attempt to hide it, and are still hiding what the firehose contained. This is not kudos-worthy.

The article was redirected to /dev/null within about two hours of publication on a Friday afternoon. We are still only about 50 hours out from that event. There have been basically zero conventional working hours since the failstorm erupted.

I am not going to say that the Ars editorial staff has necessarily covered itself in glory here--you could make an argument that this should have been an "all hands on deck, 6a-6p work, Christmas is cancelled" event--but to me it does not currently seem to be dripping with concentrated asscoverium.

 

[Sarty]

Leaving aside whether it’s doable at all, getting 99.9% reliable LLM output is the holy grail of the field and anyone doing it would be screaming it from the rooftops.

Giles Corey, screaming in his death throes: MORE NINES

 

[Sarty]

This is a collective failure, and no one person is fully responsible.

When someone steals from the company account, in a sense, the whole company failed. But one guy was the thief.

 

[Sarty]

Of all people to make this same mistake again, it is less likely to be this author. His reporting will be viewed with a critical eye for years, and he knows it. He'll be on his best behavior because Ars has proven it won't tolerate the black eye, evidence by the retraction and acknowledgement.

This is hilarious and sad. Mr. Edwards' continued employment would prove beyond a shadow of a doubt that Ars Technica is absolutely willing to tolerate this misconduct.

 

[Sarty]

While you are absolutely entitled to your opinions I find it particularly bold for non-subscribers like yourselves to disparage paying readers' opinions on this. You have no skin in this game.

You say "non-subscribers". I say "former subscribers". For a reason.

 

[Sarty]

I understand the calls to not be bloodthirsty but that's not really what's going on here in the first place and frankly bad faith to keep calling out people for that. Rather it's What are the grounds for dismissal? And specifically, for a journalist.

For those commentators who are denouncing what they see as torches and pitchforks, it's important to understand that the he-must-be-fired brigade do not necessarily see this as a question of punishment. Rather, it is: is Benj Edwards fit for purpose? Is he able to fulfill the duties and responsibilities of an Ars Technica journalist on February 17, 2026 and beyond?

These duties and responsibilities include, but are not limited to, drawing in traffic (paid and unpaid) through trustworthiness and professionalism. We've seen many readers express their concern about ending their subscriptions. Would you really bother to ever click on a Benj Edwards article again? Duties and responsibilities apparently included joint writing projects with other Ars journalists. If you worked for Ars, would you ever share a byline with Mr. Edwards?

The answers to these questions, to my mind, are an unfortunate but utterly deserved "hell no".

 

[Sarty]

Sorry how dare I try to provide a reasoned perspective instead of screaming into the void like all the people who created accounts just to complain about this.

Contrary to the usual semi-conspiratorial viewpoint on paid actors or bots, creating an account and leaving your first comment is actually rather high-friction. If somebody has never gone through the effort to create a user account and leave a comment but they specifically chose to do so for this story, the baseline expectation is that they actually felt very strongly about this, far above and beyond their reaction to any normal story they read on Ars Technica. Discounting their first comment viewpoint, purely on the basis of it being their first comment, is probably the opposite of what you should be doing.

 

[Sarty]

You'll only see this if you manically refresh the page, but what you can't see is that "KobayashiSaru" has made two different edits to comment #934, each more spittle-flecked than the last.

I don't know what kind of life experience drives such strident defense of fraudulent work behavior. It's foreign to me.

 

[Sarty]

Hopefully Benj Edwards gets "ejected" from writing AI articles for a bit. It would be absurd to hold us commentators (not least of which the legendary Jim Salter) to a higher standard than paid writers.

This combination of factors was also not lost on me. The magnitude of that potential contradiction gives me an additional degree of confidence that the final resolution of this saga will be a rather severe one.

 

[Sarty]

Everything in the article was written by humans, except the quotes

Momentarily leaving aside allllll of the other problems here, this is not how the word "everything" works!

 

[Sarty]

That was actually part of my point: it's important to be seen to be doing the right thing. As you point out, currently the idea that there will be a full and open investigation is left to be taken on faith.

Some form of "we will have more to say on this topic once we have completed our investigation, but that will take some time" should have been the last line of Ken Fisher's statement, and I don't think there was a good reason to omit it.

"We're not used to dealing with fuckups of this magnitude, and it just didn't occur to me to be that explicit" is about the best I can do.

 

[Sarty]

The long story short is that there's a whole shit-ton of stuff to unpack before decisions can be made.

Saving his colleagues the profound mess and headaches that he has caused them, when the final outcome probably is not in doubt, is one of the reasons I think Mr. Edwards should immediately resign his position (perhaps, at management's discretion, after participating in a process failure analysis that is clearly sorely needed). If he ever wants to work again in this or a related field, it would be a tiny first step towards building back up his professional reputation.

Whether I, as a hiring manager, conditioned on everything else would think about hiring somebody who owned the enormity of their fuckup, versus one who has demonstrated that they will fight tooth and claw to gouge every last drop of blood from an employer whose reputation they damaged.

(blah blah caveat about "alleged" conduct--but in this case, presumably Mr. Edwards does know what he did or didn't do)

 

[Sarty]

It's a bit deeper than that, and as much as I enjoy a good prank, consider this fair warning a bucket of cold water may be coming:

This is an excellent comment, and it really underscores just how badly Mr. Edwards fucked over his colleagues.

 

[Sarty]

Agreed. The soccer rule of two yellow cards (warnings) then a red one (out!) ought to apply.

Amusingly, this is not how soccer works. A major transgression absolutely will result in immediate ejection. In fact, a not-insignificant aspect of arguing with your buds while watching a soccer game is whether a player's first transgression merits immediate ejection or merely a caution.

And that's the whole 26 pages of comments! Does a journalist fraudulently ascribing fabricated quotes amount to an immediate-termination offense?

 

[Sarty]

They don't need to be terminated, and potentially canceled in the industry, to learn. They can be counseled and educated. They can be made to understand that this can't happen again.

You're describing how you mentor an intern, not how you discipline a senior staffer.

 

[Sarty]

Well, Soccer players given an ejection are usually ejected for the match. The system would be somewhat more controversial if ejection meant being fired and barred from taking a position at another team, per your earlier suggestion.

Maybe someone else said Mr. Edwards should be barred from the industry--I don't think I did. What I did more or less say that if he ever hopes to convince any particular journalism outlet to hire him, he has some enormous work ahead of him. No individual outlet is obligated to hire him. He doesn't have a perpetual right to a job in this field, certainly not if he is a known fraudster.

The outcome of those two concepts is probably the same--I somewhat doubt I'll ever see another Benj Edwards byline at a reputable outlet. But I think the difference is still very important.

 

[Sarty]

Now, should Benj be fired? I think he should be dismissed from the AI beat, but not dismissed from reviewing video games or classic computers or other non-AI subjects.

This one merits pondering. I might be able to live with something like that. Specifically, no articles about anything serious.

Whether it's tenable to retain a second-class reporter--someone specifically barred from actual heavy topics--is an interesting question. Does the reputational damage still attach? Is that viable from an office politics perspective? Would Mr. Edwards even be interested?

 

[Sarty]

These three things are known, and I believe they are enough to make a decision on Benj's fate. The question becomes: What does "not permit" actually mean? I'd be happy with a promise not to use AI again and a breakdown of how things will change at Ars Technica as a result.

A promise to not use AI in direct and specific violation of policy again, plus two bucks, will get me a nice coffee at Starbucks.

Maybe three dollars now, damn inflation.

 

[Sarty]

A red card does not mean booting the player from the team permanently - unless it was the result of deliberate malice.

No team - and you damn well know that - would boot a player permanently for getting a red card in a match.

If you incorrectly think that I am the one who introduced the soccer yellow/red card metaphor to describe Mr. Edwards' apparent misconduct and how Ars Technica might seek to discipline him, perhaps you unwisely used an AI tool to summarize the comment section up to that point.

 

[Sarty]

well now that they got caught they'll make extra-certain they don't get caught again

There's an old saying in Tennessee--I know it's in Texas, probably in Tennessee--that says, fool me once, shame on--shame on you. Fool me--you can't get fooled again.

 

[Sarty]

I don't believe it's likely that Mr. Edwards intended to quote the subject of his article incorrectly, therefore I don't think those adjectives accurately convey the situation.

So back to intent - does it matter? My initial reaction is that it should matter, but I'm open to other thoughts on the matter.

I think "reckless disregard for the truth" is the applicable standard here. Putative AI expert asks the slopbox to pull quotes from a single blog post, an act which would take that expert maybe five minutes of human time, and he just rolls with it and publishes it to the world.

If "did not know and had no reasonable way of knowing" is a 0, and "I lied to them intentionally, maliciously, and with a smile on my face" is a 10, Mr. Edwards' apparent behavior isn't a 10. But it's closer to a 7 than a 3.

 

[Sarty]

We're getting pretty close to the edge of inside baseball that I'm not sure I want to cross, even as a former (not current) staffer. But in the interests of transparency, I'll tell you a little about how this worked four or five years ago.

If it eases that twitch you're feeling in the back of your eye, this comment is absolutely a level of detail I regularly get into with prospective new employees during an interview. And if the tables were turned, I would be uncomfortably shifting in my seat if the hiring manager wouldn't tell me even that much.

 

[Sarty]

I have to urge more patience. Until we hear otherwise I'm prepared to give a week, but I would desperately want at least an update that something is being done.

We know this comment section is being watched, as it should be. If someone has not yet told senior staff that they are burning their accumulated trust and political capital by each passing hour--not necessarily in issuing a final report but in saying "we really are working on it"--then they are doing management (who should know better anyway) a grave disservice.

If a correct decision is made in two weeks but no guidance whatsoever is given in the interim, how many will be left to see the final report?

---
edit: which is to say I completely agree with you, in case I'm not clear in the point I'm trying to convey

 

[Sarty]

An interview is:

• under cover of commercial confidentiality

lol.

 

[Sarty]

Which is a long way of saying: those of us who grew up reading newspapers or watching nationally edited TV news prior to 2000 experienced a thoroughness and level of precision in reporting and writing that no longer exists and will probably never exist again.

Alas, probably never existed very long before, either. The age of yellow journalism was not something to celebrate.

I've seen it persuasively argued that only the centralization/consolidation of major news organizations, combined with the advent of large-format retailers that had to compete with each other and make huge local advertising purchases, made the large, professional, triple-checking newsroom a realistic possibility.

But I'm not that kind of doctor.